ELECTRICAL CONDUCTIVITY OF AIR AND SALT VAPOURS. 
423 
would drop in about one minute. If after passing this current for 10 minutes, the 
circuit was broken for 10 seconds, then, on remaking it, the initial current would be 
perhaps 20 j)er cent, greater than the final steady value to which it would fall in 
a few seconds. 
Long-continued heatiug of the tube to 1500° was found to diminish this effect, 
which, however, slowly reappeared, though not entirely, when the tube was cooled. 
It was also found that the steady currents observed through air fell off considerably 
as the platinum tubes became more and more aged through use. 
This effect is presumably due to long-continued heating of tlie platinum. A 
similar effect was observed by Elster and Geitel (‘Wied. Ann.,’ vol. 37, p. 315, 
1889) in their experiments on tlie charge communicated to a neighbouring plate by 
an incandescent platinum wire in air. At first the plate gets a positive charge, but 
this diminishes with long-continued heating of the wire, and finally at low pressures 
actually changes sign and the plate then gets a negative charge. Nahrwold (‘ Wied. 
Ann.,’ vol. 35, p. 107, 1888) ascribes the positive electrification first observed to the 
action of dust in the air, and reo-ards the final communication of a negative charo’e 
as the normal effect. 
It is clear that the falling off in the amount of conductivity as tlie tube gets 
aged is due to a change of some kind in the state of the platinum. Perhaps 
occluded gases are gradually expelled from it, though it seems very unlikely that 
the loss of a small quantity of gas should produce such a marked change in the 
amount of ionization as actually occurs. I think a change in the state of molecular 
aggregation of the platinum, due to long-continued heating, is the most probable 
cause of the falling off in the amount of ionization produced in the air close to it, 
though it is not easy to see why any such change should have this effect. Perhaps 
the porosity of the metal diminishes with long-continued heating, so that the 
effective surface of metal exposed to the gas is diminished. 
The large current obtained for a short time on first applying the E.M.F. suggests 
the idea that the ])assage of the current produces some kind of polarisation of the 
electrodes, luit on connecting the electrodes directly to the galvanometer immediately 
after applying a big E.M.F. for some time, no indication of a polarisation current 
could be detected. If the applied E.M.F. was suddenly reversed in direction, 
the current after reversing was for a short time greater than the steady value to 
which it soon settled down, but this effect was not greater than what would have 
been observed if the E.M.F. had merely been suddenly applied in the second direction 
without first applying it in the opposite direction. 
The object of the investigation being to ol)tain information on the variation of the 
conductivity with temperature, the investigation of these peculiar effects was not 
proceeded with, but I hope before long to make experiments with a view to elucidate 
their real nature. 
The values given for the observed currents through air throughout this paper are 
